A Survey of Semantic Approaches in Personality Traits Profiling Analysis

Natural language processing (NLP) and machine learning have been used in the analysis of individual profiling based on various data sources, including public datasets, social media platforms, and frameworks such as DISC, Big Five, MBTI, Holland’s RIASEC, 16 PF, HEXACO, and others. Deep learning is also utilized in research, with algorithms such as CNN, LSTM, and Galvanic Skin Response (GSR) employed for classification, labeling, and regression. The development of profiling analysis has led to a semantic approach to enhance model performance. Examples of new models that can improve performance include the Semantic-enhanced Personality Recognition Neural Network (SEPRNN) and the Robustly Optimized BERT Pre-Training Approach (RoBERTa). This study begins by collecting papers from publication databases. The next step is paper selection, and then the paper content is analyzed according to the research questions that have been determined. The final step is visualization. This study aims to provide a comprehensive understanding of the development of NLP in profiling analysis. The results of this study show that 53% used social media as a data source in profiling analysis, 43% used deep learning for the learning model, 68% used the Big-Five personality framework, 10.2% used support vector machine (SVM) as the algorithm, and 86.7% used lexical semantics for approaches. Research opportunities in profiling analysis include developing and improving predictive models, integrating and exploring NLP, using diverse data and machine learning techniques, applying them in various contexts, and integrating them with other technologies.